Native listeners more effectively utilize spatial release from masking when detecting speech in noise and reverberation

Abstract

This study examined how nativeness affects the performance of spatial release from masking (SRM) during a task detecting speech in noise under anechoic and reverberant environments. While previous research has primarily focused on speech intelligibility, the differences in SRM performance in terms of detecting speech especially between native and nonnative languages remain unclear. In the present study, native Japanese listeners performed a detection task with native (Japanese) and nonnative (English) target utterances. Experiments were conducted in anechoic and reverberant environments. The target speech was fixed at 0°, while the noise was presented from five azimuthal angles ranging from 0° to 180° in 45° steps. The results showed no significant difference between languages in the anechoic condition. However, a significant interaction was observed in the reverberant condition: the native language showed a significantly larger SRM effect than the nonnative language. Specifically, while SRM effect for the nonnative language was reduced due to reverberation, high SRM effect for the native language remained. These results suggest that the nativeness of the target speech is a critical factor for the SRM effect, specifically in reverberant environments. [Work was supported by Sophia University Special Grant for Academic Research (Research in Priority Areas).]